See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/238079823

Sampling and analysis of aerosols

Article  in  Pure and Applied Chemistry · January 1983

DOI: 10.1351/pac198355121925

CITATIONS READS

6 86

1 author:

Freddy Adams
University of Antwerp
74 PUBLICATIONS   1,008 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Freddy Adams on 02 January 2014.

The user has requested enhancement of the downloaded file.

Pure & Appi. Chem., Vol. 55, No. 12, pp. 1925—1942, 1983. 00334545/83 $3.OO+O.OO
Printed in Great Britain Pergamon Press Ltd.
©1983 IUPAC

SAMPLING AND ANALYSIS OF AEROSOLS

F. Adams

Department of Chemistry, University of Antwerpen (U.I.A.), B-2610 Wilrijk, Belgium

Abstract - Assessment of the potential environmental hazard and toxic effects of

aerosols emitted into the atmosphere requires their detailed physical and chemical
characterisation. A number of instrumental, mostly multi-element or multicompo-
nent, analytical methods are used for the sensitive bulk characterisation of par-
ticulate matter sampled from the atmosphere. Information on an individual particle
basis can also be obtained with a number of microanalytical techniques. Surface
analysis is used to gain insight into the heterogeneous distribution and the sur-
face predominance of elements or organic compounds. Some methods allow chemical

speciation.

INTRODUCT ION

Most metallic elements and a considerable number of organic compounds are emitted into the
atmosphere in association with solid or liquid particles. Certain more volatile elements
and species are emitted as vapors and the surface of finely divided particles may act as
sitesfor adsorption or condensation. On the particle surface catalytically or photo-
chemically induced reactions may further occur. The airborne particulate matter which is
hence formed directly or through secondary reactions has a widely varying composition,
aerodynamic dimension, shape and degree of uniformity. The aerosol thus consists of a
complex interacting dynamic system. Particulate matter amounts to more than 50 % of all
air pollutants.

Theassessment of the potential environmental and toxic effects of particulate matter in the
atmosphere requires their detailed physical and chemical characterisation. Of prime impor-
tance is the size of the particles, their overall and microscopical chemical composition
and finally their chemical reactivity and physical characteristics.

Methods for monitoring or studying particulate air pollution consist in general of several
stages sample collection, analysis and finally data interpretation. These distinct
stages in the entire problem solving process are interdependent. They also depend on the
overall area of concern (ambient, source or exposure monitoring), then on the analytical

entities under scrutiny (trace elements, organic compounds or specific crystallographic

structures) and finally on the aim of the study (human toxicity, pollution abatement,
meteorology or climatology).

In what follows I will review some recent advances in the field from the (biased) experience
of someone involved with environmental analytical chemistry; examples will be selected

1925
1926 F. ADAMS

predominantly from my own laboratory's experience in bulk aerosol characterisation and the
analysis of individual particles and particle surface regions.

SAMPLING

Few methods al low a direct measurement of particulate matter in the air. Different methods
of separation use sedimentation, electrostatic or inertial collection, centrifugation or
filtration. Filter sampling is often used and is still prescribed in most official stand-
ard procedures as it is easy, straightforward and cheap. The bulk aerosol collected with
filtration samplers depends, however, on their design characteristics; the cut-off dia-
meter of the larger particles is especially pooriy defined. Any discussion of environmen-
tal standards and measurements implies the use of terms such as total dust or total sus-
pended material. There are not strictly definable as the falling velocity of particulate
matter depends on the air velocity. The only practical upper limit to particle size is
that beyond a certain size particles become so rare that they make a negligible contribu-
tion to the total airborne mass. Common bulk samplers do not collect all sizes with equal
probability and accidental selection characteristics depend strongly on the speed and direc-
tion of any external wind. It is hence surprising that the bulk aerosol sampled is still
to this date so often referred to with little specification of the actual sampling condi-
tions.

Filter sampling is easily affected by various sources of errors : direct errors in air
volume or flow rate, air leaks in the sampling train, wrong calibrations, sample losses
during or after sample collection and wrong estimates of filter blanks or contamination.
Lack of agreement between identical high volume aerosol samplers to the extent of 10 %
are common (Ref. 1), thus making sampling errors often more important than those in subse-

quent analytical procedures.

Filter collection systems have no capability of resolving the particle size distribution.
This precludes measurements in terms of the "inhalable particulate matter" i.e. the fraction
of the size population likely to penetrate in the human respiratory system. Technological
advances towards an increased burning efficiency of fossil fuel and better control tech-
niques for stack emissions of particles decrease large particle emissions considerably and
hence increase the relative importance of particles in the respirable range.

Of the collection systems with a capability of size classification, inertial impactors are
the most popular. Though there are many different types the underlying principle is simple
and the same for all : an aerosol jet is forced to change its direction rapidly and par-
ticles too large for following the streamlines are collected by impingement on a collection
plate. In "virtual" impactors the same principle is used to separate particles according
to their inertia in two partial streams where they remain airborne and can be separated
e.g. by filtration. Cascade impactor arrangements use this principle repeatedly to collect
particles in several (typically 6-8) size fractions. An overview of commercially available
instruments is given by Maple and Willeke (2).
 Sampling and analysis of aerosols 1927

The cut-off sizes of any single impactor stage can be calculated accurately from the geo-
metric, fluid and particle parameters. Theory shows that properly designed impactor
stages will provide sharp classification between particles above a certain aerodynamic
diameter collected and smaller particles which are not.Special precautions must, however,
be taken to insure reliable and accurate data. Disturbing effects may be due to inter-
stage losses (particles deposited on surfaces other than the impactor plates), particle
reentrainment (bounce-off and blow-off in the air stream) and inlet losses dependent on
the air velocity at the sampling location. A practical example of the importance of par-
ticle size classification is given in Table 1, where the mass median diameter for iron,
lead and benzo[ a ]pyrene for a typical ambient aerosol is used to calculate the fraction

deposited in several respiratory regions of the lung (ref. 3).

TABLE 1. Mass median diameter of iron, lead and benzo[a}pyrene of typical urban
aerosol and percentage deposited in respiratory tract

mass median nasopharyngal, tracheo- pulmonary

diameter bronchial

iron 2.7 48 7 22
lead 0.56 17 6 32
benzo [aJ 0.15 5 7 39
pyrene

Whereas normal impactors are designed to provide a sharp cut-off, in some cases it may be
desirable for the cut-off to be a specific function of particle size. Such a design for
a cut-off efficiency simulating lung retention allows the collection of respirable dust.
The direct advantage of such a system is that only two distinct particulate fractions need
to be analysed to provide a measure of human exposure to certain air pollutants. One
practical design (Ref. 4) simulates the respirable dust curve proposed by the U.S. Confe-
rence of Governmental Industrial Hygienists and is shown in Fig. 1, as an example.

Fig. 1. Comparison of the ACGIH

respirable curve to theore-
Ui
>' tical and experimental col-
0. •
C lection efficiencies of
a)
C.)

'I- impactors (Ref. 4).

C
0
C.)
a)
0
C)

Partict.e aerodynamic dia.,Dp(m)

1928 F. ADAMS

The technical difficulties in designing systems for aerosol sampling are compounded by
lack of international agreement on what is understood by respirable dust. In fact, the
standardisation needs to reconcile different national practices and philosophies, some-
times backed by decades of practical use.Through its Technical Committee Nr. 146
(ISO TC 146), the International Standards Organisation attempts specifications that can
be applied uniformly in the workplace and in the general environment.

The measurement of particle emissions in industrial process streams e.g. in stacks, is

made with filter samplers, cascade impactors or cyclones. Problems arise from the general-
ly high velocity of the gas stream, the high temperature and the moisture content. All
this makes representative particle sampling frQm the often very hostile environment of
stacks a complicated task, requiring extensive manpower and resources. Test methods for
stationary and mobile emissions tend to be specific for each industry and have a limited
application. Also, the measurements are taken over short periods of time and because
pollutant concentrations normally fluctuate temporarily and spatially, sequential samples
cannot often be considered as representative.

AEROSOL ANALYSIS

Bulk aerosols and size fractionated particulates have in the past been performed with many
different inorgahic and organic analytical techniques. Emphasis has to be placed on rapid,
sensitive, multi-element (or for organic compounds multi-component) methods. Not only is
rapid and panoramic analysis necessary to handle the large sample loads often encountered,
but also these characteristics help to reveal significant correlations between different
pollutant components. In additiQn, the ability to measure more elements or components
than just a few pollutants, often does not significantly raise analytical costs, but
greatly increases the usefulnessof the results. Another important general factor is
specificity. There is a growing interest in analytical methods capable of identifying
the actual chemical compounds in which potentially toxic trace elements exist in the
aerosol.

Some of the more successful analytical techniques derive their advantage from the absence
ofchemical separation. For instance for inorganic analysis dissolution of the sample is
not a trivial task. Hence direct methods of analysis - that is without dissolution - have
a net advantage.

Inorganic analysis
The elemental analysis of air pollution particulate samples is a unique problem. The
total amount of material is ordinarily small but the sample may contain a large number of
elements over a broad atomic number range and at widely differing concentrations. An
effective analytical scheme must measure a considerable number of elements of interest;
it must have good detectability in order to provide reliable results at low concentrations
even in the presence of large amQunts of Qther elements. Several methods have been widely
used
-
X-ray spectrometry is attractive because it meets most of the requirements for some 20
elements. The methodology applied for air pollution analysis utilises mostly energy-
dispersive detectors and excitation with X-ray tubes (EDXRF) or protons (or other ele-
mentary ions (PIXE)). Especially in the latter excitation mode sample weight can be
 Sampling and analysis of aerosols 1929

extremely small which allows short term sampling within a reasonably short analysis time.
The difficulties inherent in X-ray spectrometry namely the data analysis of the energy
dispersive spectra, and matrix absorption of the fluorescent X-rays have gradually been
solved by computer analysis.
- Instrumental neutron activation analysis has been extensively used. A recent literature
survey lists 225 references (Ref. 5). A drawback is that toxicologically very important
elements such as lead or cadmium are insensitive.
-
Spectrometric methods of analysis suffer from the need to dissolve the particulate matter
(atomic absorption spectrometry or inductively coupled plasma emission spectrometry) or
from limitations in sensitivity or precision (d.c. are emission spectrometry).
There is evidence that the Zeeman background correction could be useful for the direct ato-
mic absorption determination of a number of elements (Ref. 6).

Organic analysis

Organic compounds reach the atmosphere from widely differing inthropogenic and natural
sources. Estimates of the direct particulate organic carbon emission range between 23 and
50 M tons g1; gas-to-particle conversion adds another 60 Mtons g1 to this number (Ref. 7).
With an average particle life time of 4-10 days this implies a total organic carbon burden
of between 1 and 6 Mton (Ref. 8). Perhaps more important is the wide variety of organic
compounds. Nevertheless the number of compounds presently determined is rather limited
in 1977 only ca. 450 had been identified in the particulate phase of the atmosphere (Ref. 9).
It is an almost hopeless task to attempt to devise methods for all of the individual organic
compounds which can be assumed to be present.

Major interest has been devoted till now to the determination of polycyclic aromatic hydro-
carbons (PAH) because this class of compounds is related to the combustion of fossil fuels
and because several of its members exhibit carcinogenic properties. Current analytical
methods are based on the separation of individual compounds by thin-layer or high perfor-
mance liquid chromatrography (TLC and HPLC) with luminescence detection, or gas chromato-
graphy with flame ionisation detection or mass spectrometry. Gas chromatography-mass spec-
trometry (GC-MS) is the method of choice. It requires first the isolation of the PAH as
a compound class followed by their specific determination. This procedure is not only time
consuming but also isomeric pairs such as benzo[a]pyrene and benzo[e]pyrene are difficult
to separate. There is a continuing interest in alternatives that would eliminate or at
least simplify the concentration steps in the procedure.

Recently alternatives have appeared luminescence using the Shpol'skii effect (Ref. 10),
matrix isolation (Ref. 11) or fluorescence line narrowing (Ref. 12) spectroscopy combined
with site selective laser excitation. Also the improved mass spectrometric identification
based on mass spectrometry-mass spectrometry (MS-MS, also called MIKES, mass analysed kine-
tic energy spectrometry) has been applied (Ref. 13).
1930 F. ADAMS

An important problem in the organic analysis with GC-MS is the lack of a reasonable mass
balance. Indeed, the method only permits the determination of relatively volatile,
non-polar and thermally stable compounds, which only account for about 50 % of the extract-
able organic matter. It has been demonstrated that PAH-derivates such a oxidised, nitrated
or sulfonated compounds are important toxicologically. Such compounds are directly
formed in car exhausts, emission stacks and plumes but especially during transport of PAH
through the atmosphere. Analysis for PAH-derivates is presently still in a qualitative
stage. Complex separations by off-line HPLC are usual ly required. Direct LC-MS coupling
or MS-MS hold promise for future work into this area.

Another group of environmentally important compounds is the chlorinated dibenzodioxins and

dibenzofurans. The different isomers of these compounds exhibit markedly different stabi-
lities and toxicities, the hexa-, penta- and octa-chloroisomers being relatively stable and
the tetra chloroisomers tetrachlorobenzodioxin (TCDD) and tetra-chlorodibenzofuran (TCDF)
extremely toxic. The dioxine isomer 2,3,7,8 TCDD Is considered the most toxic with an

acute lethality (LD 50) in guinea pigs of 1 pg kg1 body weight (Ref. 15).

In 1977 the first reports on significant amounts of dioxins adsorbed on precipitated fly
ash samples from municipal incinerators appeared. Even then, the analytical methodology
needed to deal with such complex isomeric mixtures at such trace levels is still evolving
and being perfected. The method of choice for the determination is GC-MS, but ratio-
immunoassay may be used for a preliminary screening of a large number of samples for the
presence or absence of the compounds. Extraction procedures, purification steps and
column packings have improved since Buser and Rappe (Ref. 16) first separated 2,3,7,8 TCDD
from environmental material (and from its isomers). Results obtained at different loca-
tions agree now sufficiently to give confidence in the results (Ref. 15).

Recent reviews on the analysis of air particulate matter for organic compounds are published

(Ref. 5, 14).

Data treatment

In the past decade multivariate data analysis methods such as e.g. principal component
analysis, factor analysis and cluster analysis have becomen valuable tools for data treat-
ment. Many applications of these methods fr the evaluation of pollution aerosol analyses
exist in the literature. These types of data sets have in general a very complicated
structure as many natural and anthropogenic sources contribute to the aerosol load and
composition. In contrast with this situation, at remote locations the aerosol is often
made up of only a few major contributing sources. The data hence become more easily under-
standable. Table 2 shows the results obtained by principal component analysis on a 34
sample set of trace elemental data from the air particulate matter in the altiplano in
Bolivia. Samples were taken with a 5 stage cascade impactor and 23 elements were systema-
tically determined with PIXE and XRF. The complexity of the data set prevented the eluci-
dation without resorting to artificial intelligence. Five distinct components explain
the variability of trace elemental concentrations : a soil derived component, a component
related to biological processes possibly forest fires and/or agricultural burning activities,
and three components of unknown origin labeled here as sulfur, lead and calcium because
of the predominance of these elements.
 Sampling and analysis of aerosols 1931

TABLE 2. Percentage variance in the data of the different size fractions of

remote aerosol explained by the identified components.

Variance explained by component (%)

Fraction pm Soil Sulphur Lead Calcium Burning

16-8 64.9 12.9 14.2 - -

8-4 58.9 16.5 15.8 - -
4-2 57.0 19.8 7.7 8.4 -
2-1 54.6 19.6 12.9 - -
1-0.5 31.3 27.3 8.9 7.7 18.5

Microscopical analysis

The majority of methods for the analysis of airborne particulate matter require at least
microgram or milligram amounts of material. Consequently, the analytical information
obtained is statistically representative of a large number of airborne particles collected.
In principle, the most definitive information can only be extracted from a collection of
particles by the analysis of a large number of carefully selected individual particles.
Indeed, with such data it is possible to obtain information about the sources of pollution
and their relative contributions to an ambient aerosol. Also, it is possible to investi-
gate changes in the aerosol as it travels from its source. Finally, the potential effects
of individual particles can be estimated in a direct and straightforward way.

Microscopical methods which have been used for the characterisation of micrometer sized
particles are based on the application of light microscopy (LM), electron probe microana-
lysis (EPMA) or scanning electron microscopy (SEM), ion microanalysis (IPMA) and transmis-
sion or scanning transmission electron microscopy (TEM or STEM).

Ligh microscopy is a powerful yet simple tool which by simple observation allows the charac-
terisation of a number of particles, their size distribution, shape and morphology down to
a size limit of 0.5 - 1 pm. Chemical identification is indirect and relies on a plethora
of properties more or less characteristic of a given chemical compound. These include
transparancy, colour, refractive index, birefringence, size, shape and morphology. Direct
chemical information is attainable by utilising microchemical reactions. These can, however,
hardly be used in routine applications and they often merely indicate the presence of a
particular cation or anion rather than providing information about the overall chemical
composition. Besides, not many people have the intensive background in mineralogical LM
required for the full exploitation of the method. Consequently, there is a tendency to
replace LM by other microscopical methods which enable chemical information to be obtained
from some associated spectroscopic measurement.

Analytical tools which are well suited for the analysis of individual particles include
the electron beam instruments : EPMA, SEM and TEM/STEM. The X-rays induced under electron
bombardment can be measured with an energy dispersive or wavelength dispersive X-ray spec-
trometer which forms the basis of a powerful, non-destructive multi-element analytical
1932 F. ADAMS

capability. Size and morphology can be examined using the backscattered or secondary
electrons generated by electron impact on the sample (in EPMA or SEM) or by using the
primary electrons transmitted through the specimen in TEM or STEM. All these instruments
provide similar data, the electron probe being optimized for analytical work and the
other instruments for morphological studies. TEM produces the least intense X-ray spectra
but is well adapted for the analysis of the smallest particles below 0.1 pm, and for the
exploitation of selected area electron diffraction (SAED) to derive crystallographic infor-
mation about selected particles or regions thereof. Routine studies in air particulate
research have been based on all of these methodologies (e.g. Ref. 18-21).

The manual analysis of particles with electron microscopes may be a very time consuming
process. Minicomputer based automatic image analysis methods for analytical electron
microscopy are now available (Ref. 22). They allow unaided rapid analysis of size, shape
and composition of a number of particles in sequence. Particles can alsobe automatically
classified. Fig. 2 shows one possible representation of part of the results of the ana-
lysis in less than 1 hr of 450 particles sampled near a metallurgical factory (Ref. 23).
The number of particles containing more than 5 % of an element is plotted as a function
of X-ray energy. The presence of a few particles which contain e.g. Ni become clearly iden-
tifiable within the collection and can be selected for a closer study. It is clear that
such measurements collect large amounts f information which can only be fully processed
with computer data reduction.

Fig. 2. Cumulative intensities of X-rays between 5 and 10 keV in 450 airborne particles
collected near metal refining plant.
 Sampling and analysis of aerosols 1933

More recently two new instruments which utilise laser irradiation have been developed for
the analysis of microscopical objects. The first is the laser microprobe mass analyser
(LAMMA) which vaporises and partly ionises a selected particle with dimension down to
1 pm. Mass analysis is performed in a time of flight mass spectrometer. The second
is the laser Raman microprobe of which two types of instruments are available, the Raman
microprobe developed at the National Bureau of Standards (Ref. 24) specifically for the
analysis of micrometer sized particles and the now commercially available Raman microprobe-
microscope MOLE (molecular optical laser examiner). Details on critical aspects of instru-
ment design and performance and on measurement criteria have been reported. (Ref. 25, 26).

The applications of LAMMA lie in the following areas;

-
qualitative and semiquantitative analysis for identifying and estimating both matrix
and trace level constituents
- as a means of establishing a fingerprint to recognise sources of particulates. Fig. 3
shows as an example the positive mass spectrum of an unknown atmospheric particle present
in the remote background aerosol in the Atacama desert (Chile) and the fingerprint spec-
trum of NaNO3 (Ref. 27). The negative mass spectra. provide further evidence for the
similarity in composition
- as a means for detecting inorganic and organic molecular constituents (see Fig. 5).

NciNO3

Na2
Na2NO

Fig. 3. Positive mass spectrum of a typical particle from Chapiquina (top) and
positive fingerprint mass spectrum of NaNO3 crystals (bottom).

The Raman microprobe also has unique capabilities for particle analysis. Examples of its
use are given by Etz and Blaha (28). They include the characterisation of inorganic
species, microparticulate organics, the characterisation of fly ash from oil-fired combus-
tion and the South Pole aerosol.

Single particle investigation is also possible by Auger electron spectroscopy and secondary
ion mass spectrometry with an ion microprobe or microscope. These methods will be described
later. Some characteristics of microanalytical techniques are summarised in Table 3.
1934 F. ADAMS

TABLE 3. Analytical characteristics of a number of microanalytical techniques.

Method EPMA ion micros- Auger micro Laser micro LAMMA

cope/probe analysis Raman

Resol ution

lateral, pm 1 1 0.3 0.5-1 0.5

depth , pm 1 5.i0 10 ? 0.1-0.3
Elements B-U all Li-U - all
detected

Isotopes no yes no no yes

Molecular some yes some yes yes
information
Abs. sensitivity,g 10-14_io-16 1o19 10_is io8 io_1Qio_
1
Rel. sensitivity 100 1 1000

(ppm)

Surface analysis

The inverse dependence of trace constituents on particle size (Table 1) is due to condensa-
tion onto the particle surfaces from the vapor phase. One can thus expect to find a number
of more volatile elements and compounds on particle surfaces.

The techniques which have been utilised for the analysis of surface regions are X-ray
photoelectron spectroscopy, XPS (or ESCA), Auger electron spectroscopy (AES) and secondary
ion mass spectrometry (SIMS). A more limited form of surface information is also available
with electron probe microanalysis (EPMA) although here the depth resolution obtained is

limited (1 pm). Of these techniques AES and SIMS are most useful, owing to the high sen-
sitivity and the good depth resolution resulting from etching (sputtering) with an ener-
getic ion beam. Both methods also posses microprobe capabilities with incident beam dia-
meters of the order of 1 pm or below. SIMS has an extremely high sensitivity ranging from
100 ppm to io_2 ppm depending on the element and the primary beam used, enabling deter-
minations at trace levels. XPS and AES provide information on molecular species and are
hence useful for speciation and for the study of heterogeneous reactions on particle
surfaces (Ref. 29).

For the most part applications have involved qualitative analyses since quantitative work,
although possible in principle, is far from straightforward. 1 shallnow consider micros-
copical SIMS as an example and consider some of the difficulties involved.
- difficulties are encountered in establishing even semi-quantitative depth scales. Depth
calibration is normally obtained by calibrating the rate of removal of surface material.
If this is more or less readily done with simple standard materials, with environmental
particles there are considerable difficulties in matching the surface composition of a
standard with that of a poorly defined aerosol or fly ash.
- the method is subject to several types of interferences and artifacts e.g. spectral
interferences from multiply charged and molecular ions which may obscure measurements
 Sampling and analysis of aerosols 1935

even at high mass resolution.

- volatilisation losses and migration of sample species may occur under the influence of
the primary ion beam. Also, the poor electrical conductivity of the samples gives rise
to problems.

Applications of SIMS to the study of air particulates have been few. The method has been
applied to semi-quantitative studies of the surface of coal fly ash (Ref. 29) and automo-
bile exhaust particles in conjunction with several other microanalytical and surface sensi-
tive techniques including EPMA, AES and ESCA. The results have proven to be of considerable
importance from the standpoint of environmental chemistry and demonstrated the need for
more quantitative determinations both of particle surface concentrations and of elemental

depth profiles.

Van Craen et al. (Ref. 30) achieved quantitative results by applying a matrix correction

procedurewhich corrects changes of composition with depth. Fig. 4 shows the variation
of concentration with erosion time for Na and Co in a typical steel dust sample. Compari-
son of the interior regions of the sample with SIMS and X-ray fluorescence proved the accu-
racy of the depth profiles as shown in Table 4 (Ref. 31).

TABLE 4. Comparison of bulk concentrations obtained by XRF and SIMS.

concentration, wt %

element XRF SIMS

Al 0.8 0.15 0.6 ± 0.15

K 1.0 0.2 0.34± 0.12
Ca 4.0 0.6 5.0 ± 0.3
Pb 1.4 0.2 1.8 ± 0.4
V 0.037 0.0075 0.0410 + 0.0145
Ti 0.037 0.011 0.027 ± 0.006

Co
5
0 Na
0
24 .•
.2 5 c

._, • •;;.
.3
.2
..
a)
ftS
, '. ••.•
C U
o ,
0 C
.
I I
0i
10 30 50
Erosion time(min) C 10 20 30
Erosion time(min)

Fig. 4. Concentration vs. erosion time depth profiles for Na4 and Co4 (Ref. 30).
1936 F. ADAMS

Solvent leaching can also provide insight into the surface predominance and even the asso-
ciation of elements in the surface region. Moreover it provides information on the solu-
bility of compounds present. For example although AES and SIMS indicate little surface
enrichment of iron in coal fly ash, aqueous leaching rapidly removes this element indicating
its presence in a readily soluble form.

There are few examples in the literature on the detection of the surface predominance of
organic material e.g. the polycyclic aromatic compounds. Laser microprobe mass analysis
used under laser desorption conditions appears to readily be able to detect organic impuri-
ties on single particles (Ref. 31). Fig. 5 shows the LAMMA laser desorption spectrum of a
Diesel exhaust particle. Several molecularions which can be attributed to polycyclic aromatic
compounds are present.
Contamination of asbestos fibers with organic products e.g. polycyclic aromatic compounds,
or plasticisers can equally be studied with laser desorption LAMMA. Fig. 6 shows on charac-
teristic example.

(N

m/e

Fig. 5. Positive laser desorption LAMMA spectrum of a Diesel

exhaust particle.

—
0.29 p3

0.23 p3

50 100 150 200 250 300 350 400 mle

Fig. 6. Positive laser mass spectra of benzo[a]pyrene doped

crocidolite (580 ppm) (A) and of di-(2-ethylhexyl)
phtalate (1O00 ppm)(B) as surface contaminants on
crocidolite.
 Sampling and analysis of aerosols 937

Determination of chemical form and crystallographical structure

There is a wealth of information available regarding the concentration of elements in

atmospheric particles. However, most of the element related problems facing environmental
and toxicological scientists remain unsolved becaused mostly the chemical form in which
the elements are present remains unknown. Also the toxicicity of some air pollutants such
as asbestos is related to their particular crystallographical structure.

We have seen in a preceding paragraph that microanalytical methods provide compound related
information. These methodologies however, are in general hampered by inadequate detection
limits. The following examples give further information on species selective determinations
in a few important areas.
1. The assessment of asbestos pollution requires analytical methods which provide an accu-
rate identification of fibers and a determination of fiber concentration. The problem
is compounded by the existence of different varieties (chrysotile, crocidolite, amosite,
tremolite,anthophyllite) and also by the occurrence of non-asbestos fibers (e.g. glass
fibers). Asbestos is present in the respirable fraction of the urban aerosol in concen-
trations of the order of a few ng m3, although they may bepresentat much higher concen-
tration levels in or near factories or mines. Diameters range between about 20 nm and
several hundred nm and lengths from ca. 100 nm to several pm. Presently the only success-
ful approach for their measurements in the particulate fraction of the air or water
samples consists in their identification with TEM or STEM. Single particle quantitative
X-ray analysis or at least the measurement of the ratios of elemental intensities is
required for a more or less safe identification. There is an uncertainty of identifica-
tion and a non-negligible possibility for a confusion between asbestos minerals and
others which happen to have a similar (silicate) composition. Therefore,morphologicai
characterisation and crystal structure as obtained by SAED are often considered man-

datory (Ref. 33, 34).

Such a procedure is very time consuming and it has been staled that asbestos is the only
of the 129 priority pollutants selected by the Environmental Protection Agency for which
a recommended procedure is not available (Ref. 35). As a result of systematic measure-
ments on some 30 different asbestos samples of widely differing origin,I believe that
LAMMA could provide a fast and cost effective alternative. The characteristic positive
andnegative fingerprint mass spectra provide enough information on elemental consti-
tuents but also the molecular ions are f considerable help. Fig. 7 shows schematically
the relative intensities and variation coefficients as obtained for the standard com-
pounds prepared by the International Union against Cancer (UICC) (Ref. 36). Compound
specific bulk methods of analysis such as X-ray diffraction (XR) are not in general
suitable as the concentration of asbestos in nearly all practical situations is too small.
2. The characterisation and classification of atmospheric sulfates is also a topic of con-
cern, arising from the possible association of sulfates with adverse health effects and
from their importance in the acid rain problem . Biggins and Harrison (37) have reviewed
the current state of knowledge and identified the sulfate species summarised in Table 5.
The dominant species H2504, NH4HSO4 and (NH4)2S04 are omitted from the table. The ter-
tiary nature of some compounds results from the reaction of secondary sulfate (formed
from SO2 in the atmosphere) with other compounds. The lead containing species given in
the table, result from vehicle exhausted lead halides. All these compounds were identi-
fied by XRD.
PAAC 55:12—F
1938 F. ADAMS

Fig 7 Relative intensity ratio to 24Mg for a number of

elemental constituents in positive LAMMA spectra.
SAA and SAC : South African amosite and crocidolite
(SAC : scale divided by 10), RCA and CCB : Rhodesian
(Zimbabwe) and Canadian crocidolite, FAN : Finnish
anthophyllite. One standard deviation is indicated for
repeated analysis of different fibers (Ref. 27).

TABLE 5. Sulfate species identified in the atmosphere.

Classification Compound

Primary Ca504.2H20
Primary Na2SO4
Primary Pb504
Primary PbO.PbSO4
Secondary PbSO4
Secondary (NH4) 2S4. NH4HSO4
Secondary ZnSO4. H20
Tertiary Pb504.(NH4)2S04
Tertiary 2CaSO4.(NH4)2S04
Tertiary Fe2(504)3.3(NH4)2S04
Tertiary Zn504. (NH4)2S04.6H20
 Sampling and analysis of aerosols 1939

There are many inherent difficulties in speciation of atmospheric strong acids because
they must be detected apart from the simultaneously present precursors (NOr, SO2).
Therefore it is in general necessary to combine a chosen detection system with an appro-
priate sampling and sample pretreatment procedure. A few in-situe techniques are avail-
able such as nephelometry for H2S04 (Ref. 38).
3. The speciation of metals is a difficult task. Nevertheless, pollution incidents invol-
ving mercury have shown that the total metal data are insufficient and often misleading
in assessing the potential hazards.
Until recently most metal speciation was performed with electrochemistry, especially
anodic stripping voltammetry, or chromatography with conventional detectors. Gas liquid
chromatography is suitable for the separation of volatile compounds but recently it has
been supplemented with more metal-specific detectors such as atomic absorption spectro-

metry (AAS) (Ref. 39) or microwave plasma detection (MWDP) (Ref. 40).

My laboratory has been involved during the last 6 years 'with GC-AAS for the determination
of selenium and lead compounds in the atmosphere. Chromatography combined with AAS proved
to be an extremely versatile and specific methodology. Detection limits down to ca. 100 pg
absolute or 100 pg m3 relathe could be reached for organolead species (Ref. 41). Three
methylated selenium compounds could be detected in environmental air and biomethylation
of selenium could be proven (Ref. 42). Organic lead compounds on the other hand derive
their significance from the huge amounts of tetraalkyllead (TAL) compounds consumed in
leaded gasoline. TAL and especially its degradation product trialkyllead halides (TriAL)
are much more toxic then inorganic lead. TALs are volatile and hence present in the
gaseous fraction of the aerosol, the TAL content of the particulate fraction being extre-
mely small. They can be collected on adsorbents or trapped cryogenically for analysis
(Ref. 43). The TriAL compounds and their further degradation products the dialkyllead salts
are considerably less volatile and hence cannot be determined readily by GC/AAS (Ref. 44).

LL
aD

.
V)

QW Qj L&J

.
Li
cL
-D aD
L.
0 u_i

5 10 15 20
Time (mm)

Fig. 8. Gas chromatogram with graphite furnace AAS detection of standard mixture
of TAL's and butylated TriAl and DiAL compounds.
1940 F. ADAMS

A breakthrough in the direct speciation of alkyllead compounds in environmental samples

was recently published by Chau et al. (Ref. 45). Di- and trialkyllead compounds and inor-
ganic lead are quantitatively extracted from aqueous solutions by chelation/extraction
with dithiocarbamate and benzene. The alkyllead compounds are butylated by Grignard
reagent to the tetraalkyl form, RnPbBu4_n = methyl or ethyl, Bu = butyl) and inorganic
lead to Bu4Pb, all of which are determined together with GC/AAS. TAL compounds in the
sample are also extracted and determined simultaneously. A detection limit of 0.1 pg 11
was obtained in water samples. Figure 8 shows a characteristic chromatogram obtained with
graphite furnace AAS detector. All individual species are clearly identifiable. Up to
now this method was not applied in air pollution research.

A less elegant procedure based on successive extractions and AAS was applied in our labo-
ratory for the specific determination of TriAL compounds (Ref. 46). It appeared that rain
water samples obtained in our near cities contains a concentration of these toxic compounds
at a level of 150 ng 1_i, a concentration about 1000 timeslower than that of inorganic
lead (Ref. 47). In view of the toxicity of these compounds (the LD is roughly an order
of magnitude higher than that of Pb2 in rats) and the possibility that they become accu-
mulated in the biosphere, this is significant. In one earlier study by Nielsen et al.
(Ref. 48) a considerably higher concentration of organic lead was found in the brain tissue
of city dwellers than in that of village people. Similar chemical speciation needs exist
for mercury, selenium, arsenic, chromium and perhaps several other elements.

Most of the TAL compounds in gasoline reacts during combustion with ethylene dihalide
'scavengers' with the consequent emission of the lead in an inorganic particulate form,
predominantly PbBrC1. The association of the halogens especially Br with lead has been
used for tracing automotive lead : PbBrCl should give a a Br/Pb mass ratio of 0.38. It
has been reported, however, that the lead halide is not stable in the atmosphere with a
progressive loss of the halogens in a volatile form (Ref. 49). Three compounds identified
by XRD as components of vehicle exhaust are PbBrCl, PbBrCI.2NII4C1, c-2PbBrCl NH4C1 PbSQ4
and Pb504.(NH4)2S04 (Ref. 50). Since lead represents only of the order of 1 % of the
total particulate matter in air, enrichment either by density separation or by preferential
sampling of the smallest particles which contain most of the lead was necessary.

Street dusts and roadside soils become enriched with lead with concentrations ranging
between the 1000-4000 mg kg1 near busy streets. Speciation of lead in dusts have been
carried out by Olson and Skgerboe (Ref. 51) and Biggins and Harrison (Ref. 52). Only a
minor proportion of lead appears to be in a crystalline form susceptible to XRD analysis.

CONCLUSION

The aerosol is a complex interacting system which requires the most sophisticated instru-
mental analytical techniques for its chemical and physical characterisation. Its complexi-
city implicates that it is an exciting topic of investigation in environmental analytical
chemistry. For routine control of the environment - a topic which was deliberately not
covered in this survey - other factors prevail especially the enormous complexity of data
sets generated.
 Sampling and analysis of aerosols 1941

Acknowledgment

The author is indebted to Eric Denoyer, Johan IDe Waele, Bert Raeymaekers and
Marc Van Craen for the use of experimental data. Our experimental work was
carried out under research grant 80/85-10 of the Interministrial Commission for

Science Policy, Belgium.

REFERENCES

1. P. Geladi and F. Adams, Environ. Technol. Letters, 3, 457-468 (1982).

2. V.A. Maple and K. Willeke, Atmos. Environ. 10, 891-896 (1976).
3. D.F.S. Natusch and J.R. Wallace, Science (Washington D.C.), 186, 695-699 (1974).
4. V.A. Maple, VDI-Berichte, 429, 65-72 (1982).
5. D.L. Fox and H.E. Jeffries, Anal. Chem., 53, 1R-13R (1981).
6. N. Pinta, A.M. De Kersabiec and M.L. Richard, Analusis 10, 207-215 (1982).
7. R.A. Duce, Pure Appl. Geophys. 116, 244-273 (1978).
8. J. Hahn, Ann. N.Y. Acad. Sci. 338, 359-376 (1980).
9. T.E. Graedel, Chemical Compounds in the Atmosphere, Academic Press, New York, 353 (1978).
10. Y. Yang, A.P. L'Silva and V.A. Fassel, Anal. Chem. 53, 894-899 (1981).
11. J.R. Maple, E.L. Wehry and G. Mamantov, Anal. (Them. 52, 920-924 (1980).
12. J.C. Brown, J.A. Duncamson and G.J. Small, Anal. Chem. 52, 1711-1715 (1980).
13. D. Zakett, J.D. Ciuper and R.G. Cooks, Anal. Chem. 53, 723-726 (1981).
14. K. Van Cauwenberghe and L. Van Vaeck, NATO Advanced Research Workshop, "Mobile source
emissions including polycyclic organic species", Eds. D. Rondia, M. Cooke and R.H. Haroz,
D. Reydel Publ. Co., Dordrecht, The Netherlands.
15. F.W. Karasek and F.I. Onuska, Anal. Chem. 54, 309A-324A (1982).
16. H.R. Buser and C. Rappe, Anal. Chem. 52, 2257-2262 (1980).
17. P. Van EspeI\ and F. Adams, Anal. Chim. Acta, in press.
18. W.M. Henry a K.I. Knapp, Env. Sci. Techn. 14, 4500 (1980).
19. J.A. Campbell \ J.C. Laul, K.K. Nielson and R.D. Smith, Anal. Chem. 56, 1032-1040 (1978).
20. D.G. Coles, R.C. Ragaini, J.M. Ondor, G.L. Fisher, LI. Silberman and B.A. Prentice,
Env. Sci. Techn. 13, 455-459 (1980).
21. S.J. Rothenberg, Atmos. Environm. 14, 455 (1980).
22. R.J. Lee and R.M. Fisher, Characterisation of Particles, NBS, Special Publication 533,
K.F.J. Heinrich ed. U.S. Department of Commerce, Washington D.C., 63-83 (1980).
23. B. Raeymaekers and F. Adams, unpublished results.
24. G.J. Rosasco and E.S. Etz, Res. & Devel. 28, 20-35 (1971).
25. E.S. Etz and J.J. Blaha, Microbeam Analysis, D.E. Newbury ed., San Francisco Press Inc.,
173-184 (1979).
26. E. Uenoyer, R. Van Grieken, F. Adams and D.F.S. Natusch, Anal. Chem. 54, 26A-41A (1981).
27. P. Surkyn, J. De Waele and F. Adams, J. Environ. Anal. Chem.,13, 257-274 (1983).
28. E.S. Etz and J.J. Blaha, Characterisation of Particles, NBS Special Publication 533,
K.F.J. Heinrich ed., U.S. Department of Commerce, Washington D.C., 153-197 (1980).
29. C.J. Powell, Characterisation of Particles, NBS Special Publication 533, K.F.J. Heinrich
ed., U.S. Department of Commers, Washington D.C., 131-137 (1980).
30. M. Van Craen, D.F.S. Natusch and F. Adams, Anal. Chem. 54, 1786-1792 (1982).
31. N. Van Craen, E.A. Denoyer, D.F.S. Natusch and F. Adams, Environ. Sci. Technol., in press.
32. E. Denoyer, T. Mauney, D.F.S. Natusch and F. Adams, Microbeam Analysis, K.F.J. Heinrich
ed., San Francisco Press, San Francisco, U.S.A., 191-196 (1982).
1942 F. ADAMS

33. J.C. Russ, Characterisation of particles, NBS Special Publication 533, K.F.J.
Heinrich ed., U.S. Department of Commerce, Washington D.C., 13-19 (1980).
34. C. Malami, E. Hoke and M. Grassenbauer, Microchimica Acta (Wien), I, 141-158 (1981).
35. L.M. Keith and W.A. Telliard, Environ. Sci. Techn., 13, 417-423 (1979).
36. V. Timbrell, J.C. Gilson and I. Webster, mt. J. of Cancer, 3, 406-415 (1968).
37. P.D.E. Biggins and R.M. Harrison, Atmos. Environm., 13, 1213-1216 (1979).
38. R.J. Charison, A.H. Vanderpol, D.S. Covert, A.P. Waggoner and N.C. Ahlquist, Atmos.
Environm., 8, 1257-1267 (1974).
39. Y.K. Chau, P.T.S. Wong and H. Saitoh, J. Chromatog. Sc, 14, 162-164 (1976).
40. D.C. Reamer, W.H. Zoller and T.C. OHaver, Anal. Chem., 50, 1449-1453 (1978).
41. W. De Jonghe, D. Chakraborti and F. Adams, Anal. Chim. Acta, 115, 89-101 (1980).
42. 5. Jiang, H. Robberecht and F. Adams, Atmos. Environm., 17, 111-114 (1983).
43. W.R.A. De Jonghe, D. Chakraborti and F.C. Adams, Anal. Chem., 52, 1974-1977 (1980).
44. W. De Jonghe and F. Adams, Fresenius Z. Anal. Chemie, 314,552-554 (1983).
45. Y.K. Chau, P.T.S. Wong and 0. Kramer, Anal. Chim. Acta, 146, 211-217 (1983).
46. W. De Jonghe, W. Van Mol and F. Adams, Anal. Chem., in press.
47. W. De Jonghe, W. Van Mol and F. Adams, Atmos. Environm., submitted.
48. 1, Nielsen, K.A. Jensen and P. Grandjean, Nature, 274, 602-603 (1978).
49. R.M. Harrison and W.T. Sturges, Atmos. Environm., 17, 311-328 (1983).
50. P.D.E. Biggins and R.M. Harrison, APCA Journal 29, 839-841 (1979).
51. K.W. Olson and R.K. Skogerboe, Environ. Sci. Technol., 9, 227-231 (1975).
52. P.D.E. Biggins and R,M. Harrison, Atmos. Environm., 13, 1213-1216 (1979).

View publication stats